கணக்கியல் ஆராய்ச்சியின் சர்வதேச இதழ்

சுருக்கம்

Forecasting the Number of COVID-19 Active Cases in Indonesia Using the Statistical Multilayer Perceptron Feedforwards Neural Networks

Yuyun Hidayat, Dhika Surya Pangestu*

COVID-19 was confirmed to first appear in Indonesia on March 2, 2020. Since the beginning of its emergence, the development of the number of COVID-19 cases in Indonesia has continued to increase, until 29 May 2021, there have been 1,809,926 people infected by COVID-19 with the number of active cases as many as 99,690 cases in Indonesia. The active case talks about COVID-19 patients who need medical care and is directly related to hospital capacity. Therefore the prediction of the number of active cases of COVID-19 is a strategic matter to pay attention to. In this study, active cases were predicted using the Multilayer Perceptron (MLP). The data used in this study came from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University. The data is the number of positive cases, recovered, and deaths of COVID-19 sufferers in Indonesia in the period 10 January 2020–29 May 2021. The results found, in testing period 19 September 2020 – 29 May 2021 or 37 weeks, forecasting active case using (7,10,2) MLP architecture with learning rates 0.01 provides the most accurate forecasting results compared to other window width and architectures. The means absolute percentage error (MAPE) is 5.27%, the root means square error (RMSE) is 8849.01, and the means absolute error (MAE) is 5703.59. This research is useful as a reference for the government in preparation for conditioning hospital bed capacity in the next two weeks based on accurate predictions of active cases of COVID-19 in Indonesia.